文摘
In many communication systems the noise is non-Gaussian and usually exhibits impulsive characteristics. The alpha stable distribution, which is the only distribution satisfies the generalized central-limit theorem and characterizes a range of behaviors from Gaussian to extremely impulsive signals, is appropriate to model this type of noise. Since the performance of the S-transform (ST) based frequency hopping (FH) signal parameter estimation methods, which are very effective in Gaussian noise, may deteriorate significantly in the presence of impulsive noise, two robust methods based on the ST, i.e., the fractional lower order ST (FLOST) and the iterative ST (IST), are proposed in this paper. Specifically, the FLOST is the combination of fractional lower order statistics and the ST, whereas the IST is the application of the minimax Huber's robust theory to the ST. Both robust ST algorithms exhibit higher time-frequency concentration in the presence of alpha stable noise. Simulation results show that the proposed algorithms are valid for FH signal parameter estimation, and they distinctively outperform the standard ST in impulsive noise environment.